Viewing: osd_io.c
// SPDX-License-Identifier: GPL-2.0
/*
* Copyright (c) 2009, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
* Copyright (c) 2012, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
*
* Author: Alex Zhuravlev <bzzz@whamcloud.com>
* Author: Mike Pershin <tappro@whamcloud.com>
*/
#define DEBUG_SUBSYSTEM S_OSD
#include <obd_support.h>
#include <lustre_net.h>
#include <obd.h>
#include <obd_class.h>
#include <lustre_disk.h>
#include <lustre_fid.h>
#include <lustre_quota.h>
#include "osd_internal.h"
#include <sys/dnode.h>
#include <sys/dbuf.h>
#include <sys/spa.h>
#include <sys/stat.h>
#include <sys/zap.h>
#include <sys/spa_impl.h>
#include <sys/zfs_znode.h>
#include <sys/dmu_tx.h>
#include <sys/dmu_objset.h>
#include <sys/dsl_prop.h>
#include <sys/sa_impl.h>
#include <sys/txg.h>
char osd_0copy_tag[] = "zerocopy";
static void osd_choose_next_blocksize(struct osd_object *obj,
loff_t off, ssize_t len);
static void dbuf_set_pending_evict(dmu_buf_t *db)
{
dmu_buf_impl_t *dbi = (dmu_buf_impl_t *)db;
dbi->db_pending_evict = TRUE;
}
static void record_start_io(struct osd_device *osd, int rw, int discont_pages)
{
struct brw_stats *h = &osd->od_brw_stats;
if (rw == READ) {
atomic_inc(&osd->od_r_in_flight);
lprocfs_oh_tally_pcpu(&h->bs_hist[BRW_R_RPC_HIST],
atomic_read(&osd->od_r_in_flight));
lprocfs_oh_tally_pcpu(&h->bs_hist[BRW_R_DISCONT_PAGES],
discont_pages);
} else {
atomic_inc(&osd->od_w_in_flight);
lprocfs_oh_tally_pcpu(&h->bs_hist[BRW_W_RPC_HIST],
atomic_read(&osd->od_w_in_flight));
lprocfs_oh_tally_pcpu(&h->bs_hist[BRW_W_DISCONT_PAGES],
discont_pages);
}
}
static void record_end_io(struct osd_device *osd, int rw,
unsigned long elapsed, int disksize, int npages)
{
struct brw_stats *h = &osd->od_brw_stats;
if (rw == READ)
atomic_dec(&osd->od_r_in_flight);
else
atomic_dec(&osd->od_w_in_flight);
lprocfs_oh_tally_log2_pcpu(&h->bs_hist[BRW_R_PAGES + rw], npages);
if (disksize > 0)
lprocfs_oh_tally_log2_pcpu(&h->bs_hist[BRW_R_DISK_IOSIZE + rw],
disksize);
if (elapsed)
lprocfs_oh_tally_log2_pcpu(&h->bs_hist[BRW_R_IO_TIME + rw],
elapsed);
}
static ssize_t __osd_read(const struct lu_env *env, struct dt_object *dt,
struct lu_buf *buf, loff_t *pos, size_t *size)
{
struct osd_object *obj = osd_dt_obj(dt);
uint64_t old_size;
int rc;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
read_lock(&obj->oo_attr_lock);
old_size = obj->oo_attr.la_size;
read_unlock(&obj->oo_attr_lock);
if (*pos + *size > old_size) {
if (old_size < *pos)
return 0;
*size = old_size - *pos;
}
rc = osd_dmu_read(osd_obj2dev(obj), obj->oo_dn, *pos, *size,
buf->lb_buf, DMU_READ_PREFETCH);
if (!rc) {
rc = *size;
*pos += *size;
}
return rc;
}
static ssize_t osd_read(const struct lu_env *env, struct dt_object *dt,
struct lu_buf *buf, loff_t *pos)
{
struct osd_device *osd = osd_obj2dev(osd_dt_obj(dt));
size_t size = buf->lb_len;
hrtime_t start = gethrtime();
s64 delta_ms;
int rc;
record_start_io(osd, READ, 0);
rc = __osd_read(env, dt, buf, pos, &size);
delta_ms = gethrtime() - start;
do_div(delta_ms, NSEC_PER_MSEC);
record_end_io(osd, READ, delta_ms, size, size >> PAGE_SHIFT);
return rc;
}
static inline ssize_t osd_read_no_record(const struct lu_env *env,
struct dt_object *dt,
struct lu_buf *buf, loff_t *pos)
{
size_t size = buf->lb_len;
return __osd_read(env, dt, buf, pos, &size);
}
static struct page *osd_dio_page_get(const struct lu_env *env)
{
struct osd_thread_info *oti = osd_oti_get(env);
struct page *page;
int cur;
if (unlikely(!oti->oti_dio_pages)) {
OBD_ALLOC_PTR_ARRAY_LARGE(oti->oti_dio_pages,
PTLRPC_MAX_BRW_PAGES);
if (!oti->oti_dio_pages)
return ERR_PTR(-ENOMEM);
}
LASSERT(oti->oti_dio_pages);
cur = oti->oti_dio_pages_used;
page = oti->oti_dio_pages[cur];
if (unlikely(!page)) {
LASSERT(cur < PTLRPC_MAX_BRW_PAGES);
page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
CDEBUG(D_MALLOC, "alloc page %px\n", page);
LASSERT(page);
oti->oti_dio_pages[cur] = page;
SetPagePrivate2(page);
}
oti->oti_dio_pages_used++;
RETURN(page);
}
static void osd_dio_page_put(const struct lu_env *env)
{
struct osd_thread_info *oti = osd_oti_get(env);
oti->oti_dio_pages_used--;
}
static int osd_zfs_fake_lnb(const struct lu_env *env,
struct niobuf_local *lnb, loff_t offset, ssize_t len,
int maxlnb)
{
int nrpages = 0;
while (len > 0) {
int poff = offset & (PAGE_SIZE - 1);
int plen = PAGE_SIZE - poff;
struct page *page;
if (nrpages >= maxlnb) {
break;
}
page = osd_dio_page_get(env);
if (IS_ERR(page)) {
nrpages = PTR_ERR(page);
break;
}
if (plen > len)
plen = len;
lnb->lnb_file_offset = offset;
lnb->lnb_page_offset = poff;
lnb->lnb_len = plen;
/* lnb->lnb_flags = rnb->rnb_flags; */
lnb->lnb_flags = 0;
lnb->lnb_rc = 0;
lnb->lnb_guard_rpc = 0;
lnb->lnb_guard_disk = 0;
lnb->lnb_locked = 0;
lnb->lnb_page = page;
lnb->lnb_dio = 1;
LASSERTF(plen <= len, "plen %u, len %lld\n", plen,
(long long) len);
offset += plen;
len -= plen;
lnb++;
nrpages++;
}
RETURN(nrpages);
}
static ssize_t osd_declare_write(const struct lu_env *env, struct dt_object *dt,
const struct lu_buf *buf, loff_t pos,
struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
loff_t _pos = pos, max = 0;
struct osd_thandle *oh;
uint64_t oid;
ENTRY;
oh = container_of(th, struct osd_thandle, ot_super);
/* in some cases declare can race with creation (e.g. llog)
* and we need to wait till object is initialized. notice
* LOHA_EXISTs is supposed to be the last step in the
* initialization
*/
/* size change (in dnode) will be declared by dmu_tx_hold_write() */
if (dt_object_exists(dt))
oid = obj->oo_dn->dn_object;
else
oid = DMU_NEW_OBJECT;
/* XXX: we still miss for append declaration support in ZFS
* -1 means append which is used by llog mostly, llog
* can grow upto LLOG_MIN_CHUNK_SIZE*8 records
*/
max = max_t(loff_t, 256 * 8 * LLOG_MIN_CHUNK_SIZE,
obj->oo_attr.la_size + (2 << 20));
if (pos == -1)
pos = max;
if (obj->oo_dn) {
loff_t tstart, tend, end = pos + buf->lb_len;
dmu_tx_hold_t *txh;
/* try to find a close declared window to fit/extend */
for (txh = list_head(&oh->ot_tx->tx_holds); txh != NULL;
txh = list_next(&oh->ot_tx->tx_holds, txh)) {
if (obj->oo_dn != txh->txh_dnode)
continue;
if (txh->txh_type != THT_WRITE)
continue;
/* bytes already declared in this handle */
tstart = txh->txh_arg1;
tend = txh->txh_arg1 + txh->txh_arg2;
if (pos < tstart)
tstart = pos;
if (tend < end)
tend = end;
/* if this is an append, then extend it */
if (_pos == -1 && txh->txh_arg1 == max)
tend += buf->lb_len;
/* don't let too big appends */
if (tend - tstart > 4*1024*1024)
continue;
if (pos >= tend || end <= tstart)
continue;
txh->txh_arg1 = tstart;
txh->txh_arg2 = tend - tstart;
return 0;
}
}
osd_tx_hold_write(oh->ot_tx, oid, obj->oo_dn, pos, buf->lb_len);
/* dt_declare_write() is usually called for system objects, such
* as llog or last_rcvd files. We needn't enforce quota on those
* objects, so always set the lqi_space as 0.
*/
RETURN(osd_declare_quota(env, osd, obj->oo_attr.la_uid,
obj->oo_attr.la_gid, obj->oo_attr.la_projid,
0, oh, NULL, OSD_QID_BLK));
}
static dmu_buf_t *osd_get_dbuf(struct osd_object *obj, uint64_t offset)
{
dmu_buf_t **dbs = obj->oo_dbs;
uint64_t blkid;
int i;
blkid = dbuf_whichblock(obj->oo_dn, 0, offset);
for (i = 0; i < OSD_MAX_DBUFS; i++) {
dmu_buf_impl_t *dbi = (void *)dbs[i];
if (!dbs[i])
continue;
if (dbi->db_blkid == blkid)
return dbs[i];
}
return (dmu_buf_t *)dbuf_hold(obj->oo_dn, blkid, osd_0copy_tag);
}
static void osd_put_dbuf(struct osd_object *obj, dmu_buf_t *db)
{
dmu_buf_t **dbs = obj->oo_dbs;
int i;
for (i = 0; i < OSD_MAX_DBUFS; i++) {
if (dbs[i] == db)
return;
}
/* get rid of dbuf with blkd > 0 */
for (i = 0; i < OSD_MAX_DBUFS; i++) {
if (dbs[i] == NULL) {
dbs[i] = db;
return;
}
if (dbs[i]->db_offset > 0) {
/* replace this one */
dbuf_rele((dmu_buf_impl_t *)dbs[i], osd_0copy_tag);
dbs[i] = db;
return;
}
}
LBUG();
}
static ssize_t osd_write_llog_header(struct osd_object *obj,
const struct lu_buf *buf, loff_t *pos,
struct osd_thandle *oh)
{
int bufoff, tocpy;
int len = buf->lb_len;
loff_t offset = *pos;
char *data = buf->lb_buf;
while (len > 0) {
dmu_buf_t *db = osd_get_dbuf(obj, offset);
bufoff = offset - db->db_offset;
tocpy = MIN(db->db_size - bufoff, len);
if (tocpy == db->db_size)
dmu_buf_will_fill(db, oh->ot_tx LL_BFILL);
else
dmu_buf_will_dirty(db, oh->ot_tx);
LASSERT(offset >= db->db_offset);
LASSERT(offset + tocpy <= db->db_offset + db->db_size);
(void) memcpy((char *)db->db_data + bufoff, data, tocpy);
if (tocpy == db->db_size)
dmu_buf_fill_done(db, oh->ot_tx LL_BFILL);
offset += tocpy;
data += tocpy;
len -= tocpy;
osd_put_dbuf(obj, db);
}
return 0;
}
static ssize_t osd_write(const struct lu_env *env, struct dt_object *dt,
const struct lu_buf *buf, loff_t *pos,
struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t offset = *pos;
int rc;
ENTRY;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
LASSERT(th != NULL);
oh = container_of(th, struct osd_thandle, ot_super);
down_read(&obj->oo_guard);
if (obj->oo_destroyed)
GOTO(out, rc = -ENOENT);
/* XXX: disable the optimization as it's not compatible
* with indexed llog and multiple writes a block in few
* threads */
if (fid_is_llog(lu_object_fid(&dt->do_lu)) && 0) {
osd_write_llog_header(obj, buf, pos, oh);
} else {
osd_dmu_write(osd, obj->oo_dn, offset, (uint64_t)buf->lb_len,
buf->lb_buf, oh->ot_tx);
}
write_lock(&obj->oo_attr_lock);
if (obj->oo_attr.la_size < offset + buf->lb_len) {
obj->oo_attr.la_size = offset + buf->lb_len;
write_unlock(&obj->oo_attr_lock);
/* osd_object_sa_update() will be copying directly from oo_attr
* into dbuf. any update within a single txg will copy the
* most actual
*/
rc = osd_object_sa_update(obj, SA_ZPL_SIZE(osd),
&obj->oo_attr.la_size, 8, oh);
if (unlikely(rc))
GOTO(out, rc);
} else {
write_unlock(&obj->oo_attr_lock);
}
*pos += buf->lb_len;
rc = buf->lb_len;
out:
up_read(&obj->oo_guard);
RETURN(rc);
}
/*
* XXX: for the moment I don't want to use lnb_flags for osd-internal
* purposes as it's not very well defined ...
* instead I use the lowest bit of the address so that:
* arc buffer: .lnb_data = abuf (arc we loan for write)
* dbuf buffer: .lnb_data = dbuf | 1 (dbuf we get for read)
* copy buffer: .lnb_page->mapping = obj (page we allocate for write)
*
* bzzz, to blame
*/
static int osd_bufs_put(const struct lu_env *env, struct dt_object *dt,
struct niobuf_local *lnb, int npages)
{
struct osd_thread_info *oti = osd_oti_get(env);
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
unsigned long ptr;
int i;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
for (i = 0; i < npages; i++) {
if (lnb[i].lnb_page == NULL)
continue;
if (lnb[i].lnb_dio) {
osd_dio_page_put(env);
lnb[i].lnb_dio = 0;
goto next;
}
if (lnb[i].lnb_page->mapping == (void *)obj) {
/* this is anonymous page allocated for copy-write */
lnb[i].lnb_page->mapping = NULL;
__free_page(lnb[i].lnb_page);
atomic_dec(&osd->od_zerocopy_alloc);
} else {
/* see comment in osd_bufs_get_read() */
ptr = (unsigned long)lnb[i].lnb_data;
if (ptr & 1UL) {
ptr &= ~1UL;
dmu_buf_rele((void *)ptr, osd_0copy_tag);
atomic_dec(&osd->od_zerocopy_pin);
} else if (lnb[i].lnb_data != NULL) {
int j, apages, abufsz;
abufsz = arc_buf_size(lnb[i].lnb_data);
apages = abufsz >> PAGE_SHIFT;
/* these references to pages must be invalidated
* to prevent access in osd_bufs_put()
*/
for (j = 0; j < apages; j++)
lnb[i + j].lnb_page = NULL;
dmu_return_arcbuf(lnb[i].lnb_data);
atomic_dec(&osd->od_zerocopy_loan);
}
}
next:
lnb[i].lnb_page = NULL;
lnb[i].lnb_data = NULL;
}
LASSERTF(oti->oti_dio_pages_used == 0, "%d\n", oti->oti_dio_pages_used);
return 0;
}
static inline struct page *kmem_to_page(void *addr)
{
if (is_vmalloc_addr(addr))
return vmalloc_to_page(addr);
else
return virt_to_page(addr);
}
/**
* Prepare buffers for read.
*
* The function maps the range described by \a off and \a len to \a lnb array.
* dmu_buf_hold_array_by_bonus() finds/creates appropriate ARC buffers, then
* we fill \a lnb array with the pages storing ARC buffers. Notice the current
* implementationt passes TRUE to dmu_buf_hold_array_by_bonus() to fill ARC
* buffers with actual data, I/O is done in the conext of osd_bufs_get_read().
* A better implementation would just return the buffers (potentially unfilled)
* and subsequent osd_read_prep() would do I/O for many ranges concurrently.
*
* \param[in] env environment
* \param[in] obj object
* \param[in] off offset in bytes
* \param[in] len the number of bytes to access
* \param[out] lnb array of local niobufs pointing to the buffers with data
*
* \retval 0 for success
* \retval negative error number of failure
*/
static int osd_bufs_get_read(const struct lu_env *env, struct osd_object *obj,
loff_t off, ssize_t len, struct niobuf_local *lnb,
int maxlnb)
{
struct osd_device *osd = osd_obj2dev(obj);
int rc, i, numbufs, npages = 0, drop_cache = 0;
hrtime_t start = gethrtime();
dmu_buf_t **dbp;
s64 delta_ms;
ENTRY;
record_start_io(osd, READ, 0);
if (obj->oo_attr.la_size >= osd->od_readcache_max_filesize)
drop_cache = 1;
/* grab buffers for read:
* OSD API let us to grab buffers first, then initiate IO(s)
* so that all required IOs will be done in parallel, but at the
* moment DMU doesn't provide us with a method to grab buffers.
* If we discover this is a vital for good performance we
* can get own replacement for dmu_buf_hold_array_by_bonus().
*/
while (len > 0 &&
(obj->oo_dn->dn_datablkshift != 0 ||
off < obj->oo_dn->dn_datablksz)) {
if (obj->oo_dn->dn_datablkshift == 0 &&
off + len > obj->oo_dn->dn_datablksz)
len = obj->oo_dn->dn_datablksz - off;
dbp = NULL;
if (unlikely(npages >= maxlnb))
GOTO(err, rc = -EOVERFLOW);
rc = -ll_dmu_buf_hold_array_by_bonus(&obj->oo_dn->dn_bonus->db,
off, len, TRUE, osd_0copy_tag,
&numbufs, &dbp,
DMU_READ_PREFETCH);
if (unlikely(rc))
GOTO(err, rc);
for (i = 0; i < numbufs; i++) {
int bufoff, tocpy, thispage;
void *dbf = dbp[i];
LASSERT(len > 0);
atomic_inc(&osd->od_zerocopy_pin);
bufoff = off - dbp[i]->db_offset;
tocpy = min_t(int, dbp[i]->db_size - bufoff, len);
/* kind of trick to differentiate dbuf vs. arcbuf */
LASSERT(((unsigned long)dbp[i] & 1) == 0);
dbf = (void *) ((unsigned long)dbp[i] | 1);
while (tocpy > 0) {
if (unlikely(npages >= maxlnb))
GOTO(err, rc = -EOVERFLOW);
thispage = PAGE_SIZE;
thispage -= bufoff & (PAGE_SIZE - 1);
thispage = min(tocpy, thispage);
lnb->lnb_rc = 0;
lnb->lnb_file_offset = off;
lnb->lnb_page_offset = bufoff & ~PAGE_MASK;
lnb->lnb_len = thispage;
lnb->lnb_page = kmem_to_page(dbp[i]->db_data +
bufoff);
/* mark just a single slot: we need this
* reference to dbuf to be released once
*/
lnb->lnb_data = dbf;
dbf = NULL;
tocpy -= thispage;
len -= thispage;
bufoff += thispage;
off += thispage;
npages++;
lnb++;
}
if (drop_cache)
dbuf_set_pending_evict(dbp[i]);
/* steal dbuf so dmu_buf_rele_array() can't free it */
dbp[i] = NULL;
}
dmu_buf_rele_array(dbp, numbufs, osd_0copy_tag);
}
delta_ms = gethrtime() - start;
do_div(delta_ms, NSEC_PER_MSEC);
record_end_io(osd, READ, delta_ms, npages * PAGE_SIZE, npages);
RETURN(npages);
err:
LASSERT(rc < 0);
if (dbp)
dmu_buf_rele_array(dbp, numbufs, osd_0copy_tag);
osd_bufs_put(env, &obj->oo_dt, lnb - npages, npages);
RETURN(rc);
}
static inline arc_buf_t *osd_request_arcbuf(dnode_t *dn, size_t bs)
{
arc_buf_t *abuf;
abuf = dmu_request_arcbuf(&dn->dn_bonus->db, bs);
if (unlikely(!abuf))
return ERR_PTR(-ENOMEM);
return abuf;
}
static int osd_bufs_get_write(const struct lu_env *env, struct osd_object *obj,
loff_t off, ssize_t len, struct niobuf_local *lnb,
int maxlnb)
{
struct osd_device *osd = osd_obj2dev(obj);
int poff, plen, off_in_block, sz_in_block;
int rc, i = 0, npages = 0;
dnode_t *dn = obj->oo_dn;
arc_buf_t *abuf;
uint32_t bs = dn->dn_datablksz;
ENTRY;
osd_choose_next_blocksize(obj, off, len);
/*
* currently only full blocks are subject to zerocopy approach:
* so that we're sure nobody is trying to update the same block
*/
while (len > 0) {
if (unlikely(npages >= maxlnb))
GOTO(out_err, rc = -EOVERFLOW);
off_in_block = off & (bs - 1);
sz_in_block = min_t(int, bs - off_in_block, len);
abuf = NULL;
if (sz_in_block == bs) {
/* full block, try to use zerocopy */
abuf = osd_request_arcbuf(dn, bs);
if (unlikely(IS_ERR(abuf)))
GOTO(out_err, rc = PTR_ERR(abuf));
}
if (abuf != NULL) {
atomic_inc(&osd->od_zerocopy_loan);
/* go over pages arcbuf contains, put them as
* local niobufs for ptlrpc's bulks
*/
while (sz_in_block > 0) {
plen = min_t(int, sz_in_block, PAGE_SIZE);
if (unlikely(npages >= maxlnb))
GOTO(out_err, rc = -EOVERFLOW);
lnb[i].lnb_file_offset = off;
lnb[i].lnb_page_offset = 0;
lnb[i].lnb_len = plen;
lnb[i].lnb_rc = 0;
if (sz_in_block == bs)
lnb[i].lnb_data = abuf;
else
lnb[i].lnb_data = NULL;
/* this one is not supposed to fail */
lnb[i].lnb_page = kmem_to_page(abuf->b_data +
off_in_block);
LASSERT(lnb[i].lnb_page);
lprocfs_counter_add(osd->od_stats,
LPROC_OSD_ZEROCOPY_IO, 1);
sz_in_block -= plen;
len -= plen;
off += plen;
off_in_block += plen;
i++;
npages++;
}
} else {
if (off_in_block == 0 && len < bs &&
off + len >= obj->oo_attr.la_size)
lprocfs_counter_add(osd->od_stats,
LPROC_OSD_TAIL_IO, 1);
/* can't use zerocopy, allocate temp. buffers */
poff = off & (PAGE_SIZE - 1);
while (sz_in_block > 0) {
plen = min_t(int, poff + sz_in_block,
PAGE_SIZE);
plen -= poff;
if (unlikely(npages >= maxlnb))
GOTO(out_err, rc = -EOVERFLOW);
lnb[i].lnb_file_offset = off;
lnb[i].lnb_page_offset = poff;
poff = 0;
lnb[i].lnb_len = plen;
lnb[i].lnb_rc = 0;
lnb[i].lnb_data = NULL;
lnb[i].lnb_page = alloc_page(OSD_GFP_IO);
if (unlikely(lnb[i].lnb_page == NULL))
GOTO(out_err, rc = -ENOMEM);
LASSERT(lnb[i].lnb_page->mapping == NULL);
lnb[i].lnb_page->mapping = (void *)obj;
atomic_inc(&osd->od_zerocopy_alloc);
lprocfs_counter_add(osd->od_stats,
LPROC_OSD_COPY_IO, 1);
sz_in_block -= plen;
len -= plen;
off += plen;
i++;
npages++;
}
}
}
RETURN(npages);
out_err:
osd_bufs_put(env, &obj->oo_dt, lnb, npages);
RETURN(rc);
}
static int osd_bufs_get(const struct lu_env *env, struct dt_object *dt,
loff_t offset, ssize_t len, struct niobuf_local *lnb,
int maxlnb, enum dt_bufs_type rw)
{
struct osd_object *obj = osd_dt_obj(dt);
int rc;
down_read(&obj->oo_guard);
if (unlikely(!dt_object_exists(dt) || obj->oo_destroyed))
GOTO(out, rc = -ENOENT);
if (CFS_FAIL_CHECK_QUIET(OBD_FAIL_OST_FAKE_RW)) {
rc = osd_zfs_fake_lnb(env, lnb, offset, len, maxlnb);
if (rc < 0)
osd_bufs_put(env, dt, lnb, maxlnb);
goto out;
}
if (rw & DT_BUFS_TYPE_WRITE)
rc = osd_bufs_get_write(env, obj, offset, len, lnb, maxlnb);
else
rc = osd_bufs_get_read(env, obj, offset, len, lnb, maxlnb);
out:
up_read(&obj->oo_guard);
return rc;
}
static int osd_write_prep(const struct lu_env *env, struct dt_object *dt,
struct niobuf_local *lnb, int npages)
{
struct osd_object *obj = osd_dt_obj(dt);
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
return 0;
}
static inline uint64_t osd_roundup2blocksz(uint64_t size,
uint64_t offset,
uint32_t blksz)
{
LASSERT(blksz > 0);
size += offset % blksz;
if (likely(is_power_of_2(blksz)))
return round_up(size, blksz);
else
return DIV_ROUND_UP_ULL(size, blksz) * blksz;
}
static int osd_declare_write_commit(const struct lu_env *env,
struct dt_object *dt,
struct niobuf_local *lnb, int npages,
struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t offset = 0;
uint32_t size = 0;
uint32_t blksz = obj->oo_dn->dn_datablksz;
int i, rc;
bool synced = false;
long long space = 0;
pgoff_t last_index = 0;
unsigned long discont_pages = 0;
enum osd_quota_local_flags local_flags = 0;
enum osd_qid_declare_flags declare_flags = OSD_QID_BLK;
ENTRY;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
LASSERT(lnb);
LASSERT(npages > 0);
oh = container_of(th, struct osd_thandle, ot_super);
for (i = 0; i < npages; i++) {
if (i > 0 && (lnb[i].lnb_file_offset >> PAGE_SHIFT) != last_index + 1)
++discont_pages;
last_index = lnb[i].lnb_file_offset >> PAGE_SHIFT;
if (lnb[i].lnb_rc)
/* ENOSPC, network RPC error, etc.
* We don't want to book space for pages which will be
* skipped in osd_write_commit(). Hence we skip pages
* with lnb_rc != 0 here too
*/
continue;
/* ignore quota for the whole request if any page is from
* client cache or written by root.
*
* XXX we could handle this on per-lnb basis as done by
* grant.
*/
if ((lnb[i].lnb_flags & OBD_BRW_NOQUOTA) ||
(lnb[i].lnb_flags & OBD_BRW_SYS_RESOURCE) ||
!(lnb[i].lnb_flags & OBD_BRW_SYNC))
declare_flags |= OSD_QID_FORCE;
/* ASYNC means that the page comes from the cache - it must be
* written anyway.
*/
if (lnb[i].lnb_flags & OBD_BRW_ASYNC)
declare_flags |= OSD_QID_IGNORE_ROOT_PRJ;
if (size == 0) {
/* first valid lnb */
offset = lnb[i].lnb_file_offset;
size = lnb[i].lnb_len;
continue;
}
if (offset + size == lnb[i].lnb_file_offset) {
/* this lnb is contiguous to the previous one */
size += lnb[i].lnb_len;
continue;
}
osd_tx_hold_write(oh->ot_tx, obj->oo_dn->dn_object,
obj->oo_dn, offset, size);
/* Estimating space to be consumed by a write is rather
* complicated with ZFS. As a consequence, we don't account for
* indirect blocks and just use as a rough estimate the worse
* case where the old space is being held by a snapshot. Quota
* overrun will be adjusted once the operation is committed, if
* required.
*/
space += osd_roundup2blocksz(size, offset, blksz);
offset = lnb[i].lnb_file_offset;
size = lnb[i].lnb_len;
}
if (size) {
osd_tx_hold_write(oh->ot_tx, obj->oo_dn->dn_object, obj->oo_dn,
offset, size);
space += osd_roundup2blocksz(size, offset, blksz);
}
/* backend zfs FS might be configured to store multiple data copies */
space *= osd->od_os->os_copies;
space = stoqb(space);
CDEBUG(D_QUOTA, "writing %d pages, reserving %lldK of quota space\n",
npages, space);
record_start_io(osd, WRITE, discont_pages);
retry:
/* acquire quota space if needed */
rc = osd_declare_quota(env, osd, obj->oo_attr.la_uid,
obj->oo_attr.la_gid, obj->oo_attr.la_projid,
space, oh, &local_flags, declare_flags);
if (!synced && rc == -EDQUOT &&
(local_flags & QUOTA_FL_SYNC) != 0) {
dt_sync(env, th->th_dev);
synced = true;
CDEBUG(D_QUOTA, "retry after sync\n");
local_flags = 0;
goto retry;
}
/* we need only to store the overquota flags in the first lnb for
* now, once we support multiple objects BRW, this code needs be
* revised.
*/
if (local_flags & QUOTA_FL_OVER_USRQUOTA)
lnb[0].lnb_flags |= OBD_BRW_OVER_USRQUOTA;
if (local_flags & QUOTA_FL_OVER_GRPQUOTA)
lnb[0].lnb_flags |= OBD_BRW_OVER_GRPQUOTA;
#ifdef ZFS_PROJINHERIT
if (local_flags & QUOTA_FL_OVER_PRJQUOTA)
lnb[0].lnb_flags |= OBD_BRW_OVER_PRJQUOTA;
if (local_flags & QUOTA_FL_ROOT_PRJQUOTA)
lnb[0].lnb_flags |= OBD_BRW_ROOT_PRJQUOTA;
#endif
RETURN(rc);
}
/**
* Policy to grow ZFS block size by write pattern.
* For sequential write, it grows block size gradually until it reaches the
* maximum blocksize the dataset can support. Otherwise, it will pick a
* a block size by the writing region of this I/O.
*/
static int osd_grow_blocksize(struct osd_object *obj, struct osd_thandle *oh)
{
struct osd_device *osd = osd_obj2dev(obj);
dnode_t *dn = obj->oo_dn;
int rc = 0;
ENTRY;
if (obj->oo_next_blocksize == 0)
return 0;
if (dn->dn_maxblkid > 0) /* can't change block size */
GOTO(out, rc);
if (dn->dn_datablksz >= osd->od_max_blksz)
GOTO(out, rc);
if (dn->dn_datablksz == obj->oo_next_blocksize)
GOTO(out, rc);
down_write(&obj->oo_guard);
if (dn->dn_datablksz < obj->oo_next_blocksize) {
CDEBUG(D_INODE, "set blksz to %u\n", obj->oo_next_blocksize);
rc = -dmu_object_set_blocksize(osd->od_os, dn->dn_object,
obj->oo_next_blocksize, 0,
oh->ot_tx);
if (rc < 0)
CDEBUG(D_ERROR, "object "DFID": change block size"
"%u -> %u error rc = %d\n",
PFID(lu_object_fid(&obj->oo_dt.do_lu)),
dn->dn_datablksz, obj->oo_next_blocksize, rc);
}
EXIT;
up_write(&obj->oo_guard);
out:
return rc;
}
static void osd_choose_next_blocksize(struct osd_object *obj,
loff_t off, ssize_t len)
{
struct osd_device *osd = osd_obj2dev(obj);
dnode_t *dn = obj->oo_dn;
uint32_t blksz;
if (dn->dn_maxblkid > 0)
return;
if (dn->dn_datablksz >= osd->od_max_blksz)
return;
if (off == obj->oo_attr.la_size)
blksz = (uint32_t)max_t(uint64_t, osd->od_min_blksz, off + len);
else
blksz = (uint32_t)max_t(uint64_t, osd->od_min_blksz, len);
blksz = (uint32_t)min_t(uint64_t, osd->od_max_blksz, blksz);
if (!is_power_of_2(blksz))
blksz = size_roundup_power2(blksz);
/* XXX: locking? */
if (blksz > obj->oo_next_blocksize)
obj->oo_next_blocksize = blksz;
}
static void osd_evict_dbufs_after_write(struct osd_object *obj,
loff_t off, ssize_t len)
{
dmu_buf_t **dbp;
int i, rc, numbufs;
rc = -ll_dmu_buf_hold_array_by_bonus(&obj->oo_dn->dn_bonus->db, off,
len, TRUE, osd_0copy_tag, &numbufs,
&dbp, DMU_READ_PREFETCH);
if (unlikely(rc))
return;
for (i = 0; i < numbufs; i++)
dbuf_set_pending_evict(dbp[i]);
dmu_buf_rele_array(dbp, numbufs, osd_0copy_tag);
}
static int osd_write_commit(const struct lu_env *env, struct dt_object *dt,
struct niobuf_local *lnb, int npages,
struct thandle *th, __u64 user_size)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
uint64_t new_size = 0;
int i, abufsz, rc = 0, drop_cache = 0;
unsigned long iosize = 0;
ENTRY;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
LASSERT(th != NULL);
oh = container_of(th, struct osd_thandle, ot_super);
/* adjust block size. Assume the buffers are sorted. */
(void)osd_grow_blocksize(obj, oh);
if (obj->oo_attr.la_size >= osd->od_readcache_max_filesize ||
lnb[npages - 1].lnb_file_offset + lnb[npages - 1].lnb_len >=
osd->od_readcache_max_filesize)
drop_cache = 1;
if (CFS_FAIL_CHECK(OBD_FAIL_OST_MAPBLK_ENOSPC))
RETURN(-ENOSPC);
/* if la_size is already bigger than specified user_size,
* ignore user_size
*/
if (obj->oo_attr.la_size > user_size)
user_size = 0;
/* LU-8791: take oo_guard to avoid the deadlock that changing block
* size and assigning arcbuf take place at the same time.
*
* Thread 1:
* osd_write_commit()
* -> osd_grow_blocksize() with osd_object::oo_guard held
* -> dmu_object_set_blocksize()
* -> dnode_set_blksz(), with dnode_t::dn_struct_rwlock
* write lock held
* -> dbuf_new_size()
* -> dmu_buf_will_dirty()
* -> dbuf_read()
* -> wait for the dbuf state to change
* Thread 2:
* osd_write_commit()
* -> dmu_assign_arcbuf()
* -> dbuf_assign_arcbuf(), set dbuf state to DB_FILL
* -> dbuf_dirty()
* -> try to hold the read lock of dnode_t::dn_struct_rwlock
*
* By taking the read lock, it can avoid thread 2 to enter into the
* critical section of assigning the arcbuf, while thread 1 is
* changing the block size.
*/
down_read(&obj->oo_guard);
if (obj->oo_destroyed) {
up_read(&obj->oo_guard);
RETURN(-ENOENT);
}
for (i = 0; i < npages; i++) {
CDEBUG(D_INODE, "write %u bytes at %u\n",
(unsigned int) lnb[i].lnb_len,
(unsigned int) lnb[i].lnb_file_offset);
if (lnb[i].lnb_rc) {
/* ENOSPC, network RPC error, etc.
* Unlike ldiskfs, zfs allocates new blocks on rewrite,
* so we skip this page if lnb_rc is set to -ENOSPC
*/
CDEBUG(D_INODE, "obj "DFID": skipping lnb[%u]: rc=%d\n",
PFID(lu_object_fid(&dt->do_lu)), i,
lnb[i].lnb_rc);
continue;
}
if (new_size < lnb[i].lnb_file_offset + lnb[i].lnb_len)
new_size = lnb[i].lnb_file_offset + lnb[i].lnb_len;
if (lnb[i].lnb_page == NULL)
continue;
if (lnb[i].lnb_page->mapping == (void *)obj) {
void *addr = kmap(lnb[i].lnb_page);
osd_dmu_write(osd, obj->oo_dn, lnb[i].lnb_file_offset,
lnb[i].lnb_len, addr +
lnb[i].lnb_page_offset, oh->ot_tx);
kunmap(kmap_to_page(addr));
iosize += lnb[i].lnb_len;
abufsz = lnb[i].lnb_len; /* to drop cache below */
} else if (lnb[i].lnb_data) {
int j, apages;
LASSERT(((unsigned long)lnb[i].lnb_data & 1) == 0);
/* buffer loaned for zerocopy, try to use it.
* notice that dmu_assign_arcbuf() is smart
* enough to recognize changed blocksize
* in this case it fallbacks to dmu_write()
*/
abufsz = arc_buf_size(lnb[i].lnb_data);
LASSERT(abufsz & PAGE_MASK);
apages = abufsz >> PAGE_SHIFT;
LASSERT(i + apages <= npages);
/* these references to pages must be invalidated
* to prevent access in osd_bufs_put()
*/
for (j = 0; j < apages; j++)
lnb[i + j].lnb_page = NULL;
ll_dmu_assign_arcbuf_by_dbuf(&obj->oo_dn->dn_bonus->db,
lnb[i].lnb_file_offset,
lnb[i].lnb_data,
oh->ot_tx,
DB_RF_MUST_SUCCEED);
/* drop the reference, otherwise osd_put_bufs()
* will be releasing it - bad!
*/
lnb[i].lnb_data = NULL;
atomic_dec(&osd->od_zerocopy_loan);
iosize += abufsz;
} else {
/* we don't want to deal with cache if nothing
* has been send to ZFS at this step
*/
continue;
}
if (!drop_cache)
continue;
/* we have to mark dbufs for eviction here because
* dmu_assign_arcbuf() may create a new dbuf for
* loaned abuf
*/
osd_evict_dbufs_after_write(obj, lnb[i].lnb_file_offset,
abufsz);
}
if (unlikely(new_size == 0)) {
/* no pages to write, no transno is needed */
th->th_local = 1;
/* it is important to return 0 even when all lnb_rc == -ENOSPC
* since ofd_commitrw_write() retries several times on ENOSPC
*/
up_read(&obj->oo_guard);
record_end_io(osd, WRITE, 0, 0, 0);
RETURN(0);
}
/* if file has grown, take user_size into account */
if (user_size && new_size > user_size)
new_size = user_size;
write_lock(&obj->oo_attr_lock);
if (obj->oo_attr.la_size < new_size) {
obj->oo_attr.la_size = new_size;
write_unlock(&obj->oo_attr_lock);
/* osd_object_sa_update() will be copying directly from
* oo_attr into dbuf. any update within a single txg will copy
* the most actual
*/
rc = osd_object_sa_update(obj, SA_ZPL_SIZE(osd),
&obj->oo_attr.la_size, 8, oh);
} else {
write_unlock(&obj->oo_attr_lock);
}
up_read(&obj->oo_guard);
record_end_io(osd, WRITE, 0, iosize, npages);
RETURN(rc);
}
static int osd_read_prep(const struct lu_env *env, struct dt_object *dt,
struct niobuf_local *lnb, int npages)
{
struct osd_object *obj = osd_dt_obj(dt);
int i;
loff_t eof;
LASSERT(dt_object_exists(dt));
LASSERT(obj->oo_dn);
read_lock(&obj->oo_attr_lock);
eof = obj->oo_attr.la_size;
read_unlock(&obj->oo_attr_lock);
for (i = 0; i < npages; i++) {
if (unlikely(lnb[i].lnb_rc < 0))
continue;
lnb[i].lnb_rc = lnb[i].lnb_len;
if (lnb[i].lnb_file_offset + lnb[i].lnb_len >= eof) {
/* send complete pages all the time */
if (eof <= lnb[i].lnb_file_offset)
lnb[i].lnb_rc = 0;
/* all subsequent rc should be 0 */
while (++i < npages)
lnb[i].lnb_rc = 0;
break;
}
}
return 0;
}
/*
* Punch/truncate an object
*
* IN: db - dmu_buf of the object to free data in.
* off - start of section to free.
* len - length of section to free (DMU_OBJECT_END => to EOF).
*
* RETURN: 0 if success
* error code if failure
*
* The transaction passed to this routine must have
* dmu_tx_hold_sa() and if off < size, dmu_tx_hold_free()
* called and then assigned to a transaction group.
*/
static int __osd_object_punch(struct osd_object *obj, objset_t *os,
dmu_tx_t *tx, uint64_t off, uint64_t len)
{
dnode_t *dn = obj->oo_dn;
uint64_t size = obj->oo_attr.la_size;
int rc = 0;
/* Confirm if transaction has been assigned to a transaction group */
LASSERT(tx->tx_txg != 0);
/* Nothing to do if file already at desired length. */
if (len == DMU_OBJECT_END && size == off)
return 0;
/* if object holds encrypted content, we need to make sure we truncate
* on an encryption unit boundary, or subsequent reads will get
* corrupted content
*/
if (len != DMU_OBJECT_END)
len -= LUSTRE_ENCRYPTION_UNIT_SIZE -
(off & ~LUSTRE_ENCRYPTION_MASK);
if (obj->oo_lma_flags & LUSTRE_ENCRYPT_FL &&
off & ~LUSTRE_ENCRYPTION_MASK)
off = (off & LUSTRE_ENCRYPTION_MASK) +
LUSTRE_ENCRYPTION_UNIT_SIZE;
/* XXX: dnode_free_range() can be used to save on dnode lookup */
if (off < size)
dmu_free_range(os, dn->dn_object, off, len, tx);
return rc;
}
static int osd_punch(const struct lu_env *env, struct dt_object *dt,
__u64 start, __u64 end, struct thandle *th)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
__u64 len;
int rc = 0;
ENTRY;
LASSERT(dt_object_exists(dt));
LASSERT(osd_invariant(obj));
LASSERT(th != NULL);
oh = container_of(th, struct osd_thandle, ot_super);
write_lock(&obj->oo_attr_lock);
/* truncate */
if (end == OBD_OBJECT_EOF || end >= obj->oo_attr.la_size)
len = DMU_OBJECT_END;
else
len = end - start;
write_unlock(&obj->oo_attr_lock);
down_read(&obj->oo_guard);
if (obj->oo_destroyed)
GOTO(out, rc = -ENOENT);
rc = __osd_object_punch(obj, osd->od_os, oh->ot_tx, start, len);
/* set new size */
if (len == DMU_OBJECT_END) {
write_lock(&obj->oo_attr_lock);
obj->oo_attr.la_size = start;
write_unlock(&obj->oo_attr_lock);
rc = osd_object_sa_update(obj, SA_ZPL_SIZE(osd),
&obj->oo_attr.la_size, 8, oh);
}
out:
up_read(&obj->oo_guard);
RETURN(rc);
}
static int osd_declare_punch(const struct lu_env *env, struct dt_object *dt,
__u64 start, __u64 end, struct thandle *handle)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
struct osd_thandle *oh;
__u64 len;
ENTRY;
oh = container_of(handle, struct osd_thandle, ot_super);
read_lock(&obj->oo_attr_lock);
if (end == OBD_OBJECT_EOF || end >= obj->oo_attr.la_size)
len = DMU_OBJECT_END;
else
len = end - start;
/* declare we'll free some blocks ... */
/* if object holds encrypted content, we need to make sure we truncate
* on an encryption unit boundary, or subsequent reads will get
* corrupted content
*/
if (obj->oo_lma_flags & LUSTRE_ENCRYPT_FL &&
start & ~LUSTRE_ENCRYPTION_MASK)
start = (start & LUSTRE_ENCRYPTION_MASK) +
LUSTRE_ENCRYPTION_UNIT_SIZE;
if (start < obj->oo_attr.la_size) {
read_unlock(&obj->oo_attr_lock);
dmu_tx_mark_netfree(oh->ot_tx);
dmu_tx_hold_free(oh->ot_tx, obj->oo_dn->dn_object, start, len);
} else {
read_unlock(&obj->oo_attr_lock);
}
RETURN(osd_declare_quota(env, osd, obj->oo_attr.la_uid,
obj->oo_attr.la_gid, obj->oo_attr.la_projid,
0, oh, NULL, OSD_QID_BLK));
}
static loff_t osd_lseek(const struct lu_env *env, struct dt_object *dt,
loff_t offset, int whence)
{
struct osd_object *obj = osd_dt_obj(dt);
struct osd_device *osd = osd_obj2dev(obj);
uint64_t size = obj->oo_attr.la_size;
uint64_t result = offset;
int rc;
boolean_t hole = whence == SEEK_HOLE;
ENTRY;
LASSERT(dt_object_exists(dt));
LASSERT(osd_invariant(obj));
LASSERT(offset >= 0);
/* for SEEK_HOLE treat 'offset' beyond the end of file as in real
* hole. LOV to decide after all if that real hole or not.
*/
if (offset >= size)
RETURN(hole ? offset : -ENXIO);
/* Currently ZFS reports no valid DATA offset if object has dirty data
* and we cannot just switch to generic way with reporting DATA on all
* file offsets and HOLE beyond end of file, because we may get HOLE
* reported correctly at some offset inside file then DATA will find
* dirty state and be reported also at that offset by generic approach.
* This is because for HOLE report ZFS doesn't check dirty state but
* does for DATA.
* The only way to get reliable results is to call txg_wait_synced()
* when ZFS reports EBUSY result and repeat lseek call and that is
* controlled via od_sync_on_lseek option.
*/
if (!osd->od_sync_on_lseek)
result = hole ? size : offset;
again:
rc = osd_dmu_offset_next(osd->od_os, obj->oo_dn->dn_object, hole,
&result);
/* dirty inode, lseek result is unreliable without sync */
if (rc == EBUSY) {
txg_wait_synced(dmu_objset_pool(osd->od_os), 0ULL);
goto again;
}
if (rc == ESRCH)
RETURN(-ENXIO);
/* ZFS is not exported all needed function, so fall back to the
* generic logic: for HOLE return file size, for DATA return
* the current offset
*/
if (rc == EOPNOTSUPP)
result = hole ? size : offset;
else if (rc)
return -rc;
/* dmu_offset_next() only works on whole blocks so may return SEEK_HOLE
* result as end of the last block instead of logical EOF which we need
*/
if (result > size)
result = size;
RETURN(result);
}
const struct dt_body_operations osd_body_ops = {
.dbo_read = osd_read,
.dbo_declare_write = osd_declare_write,
.dbo_write = osd_write,
.dbo_bufs_get = osd_bufs_get,
.dbo_bufs_put = osd_bufs_put,
.dbo_write_prep = osd_write_prep,
.dbo_declare_write_commit = osd_declare_write_commit,
.dbo_write_commit = osd_write_commit,
.dbo_read_prep = osd_read_prep,
.dbo_declare_punch = osd_declare_punch,
.dbo_punch = osd_punch,
.dbo_lseek = osd_lseek,
};
const struct dt_body_operations osd_body_scrub_ops = {
.dbo_read = osd_read_no_record,
.dbo_declare_write = osd_declare_write,
.dbo_write = osd_write,
};
